Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Update oss_sdp_fsdp.rst #1122

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update oss_sdp_fsdp.rst #1122

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/source/deep_dive/oss_sdp_fsdp.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,7 +149,7 @@ for a shard of the model parameters. FairScale implements parameter sharding by
Data Parallel (FSDP) API which is heavily inspired by `ZeRO-3 <https://arxiv.org/pdf/1910.02054.pdf>`_. Parameter sharding is possible because of
two key insights:

1. The allreduce operation can be broken up into reduce and allgather similar to the previous sharding
1. The allreduce operation can be broken up into reducescatter and allgather similar to the previous sharding
technologies (optimizer state and gradient).

2. Individual layers can be wrapped with the FSDP API that allows us to bring in all the parameters
Expand Down